1. No category

An interface model of phonologically determined
agreement
Hannah Sande
March 29, 2015
1
Introduction
Goals:
1. Describe the typologically remarkable phonologically determined agreement system
of Gu´ebie, an undescribed Kru language (Niger-Congo) spoken in Cˆote d’Ivoire.
2. Determine an ideal model to account for this and other cases of phonologically
determined agreement.
Phonologically determined agreement systems challenge our assumption that
syntax cannot access phonological features (cf. Pullum and Zwicky 1986, 1988).
Syntax
Phonological Form
Logical Form
The big question: Can we model these systems without requiring
phonological features to be present in the syntax?
The data:
• The data here comes from original work with Gu´ebie speakers in Berkeley and
Gnagbodougnoa, Cˆote d’Ivoire from September 2013 through July 2014.
– Six speakers, ages 19-76
– One woman, five men
– Combination of text and elicitation
2
Phonological agreement in Gu´
ebie
2.1
Phonological agreement between pronouns and antecedent
nominals
• Human third person pronouns take the form /O3 /, singular, and /wa3 /, plural.
• Non-human third person pronouns agree with their nominal antecedent not in semantic class, but in phonological features.
An interface model of phonologically determined agreement
Hannah Sande, UC Berkeley
WCCFL
March 2015
– The final vowel of the noun stem determines the vowel of the pronoun.
(1) Mapping of
Final vowel
i, I, e, E
@, a
u, U, o, O
Gu´
ebie root-final vowels to pronoun vowels
3.sg pronoun vowel Plural suffix
3.pl pronoun vowel
→
e
-i
→
i
→
@
-a
→
wa
→
u
• The complete personal pronoun chart is given in (2).
(2) Human and non-human subject pronouns
Human
Non-human
Singular Plural
Singular Plural
4
3
1st e
a
1st –
–
2nd e2
a2
2nd –
–
3
3
3 3 3
3rd O
wa
3rd e ,@ ,u
i3 ,wa3
• Non-human pronouns are always phonologically determined by their antecedents.
(3) Phonological agreement of pronouns with antecedents
Noun
a. éie2.2
b. kw ala4.2
c. to3
Gloss
‘a prison’
‘a farm’
‘battle’
Object
e-4 ni-4 e2 ji3
e-4 ni-4 @2 ji3
e-4 ni-4 u2 ji3
Gloss
‘I see it (prison)’
‘I saw it (farm)
‘I saw it (battle)’
Subject
e-3 kadE3.2
@-3 kadE3.2
u-3 kadE3.2
Gloss
‘It (prison) is big.’
‘It (farm) is big.’
‘It (battle) is big.’
• This process is completely productive and without exception.
– It holds in loan words and nonce words.
(4) Phonological agreement in loan words from English/French
ji3
a. sukulu1.1.3 kOda.3.21 e-4 ni-4 u2
school
exist.
I see it(school).acc see
‘There is a school. I saw it(the school).’
ji3
b. baraZE2.3.2 kOda.3.21 e-4 ni-4 e2
I see it(dam).acc see
exist.
dam
‘There is a dam. I saw it(the dam)’
(5) Phonological agreement in nonce words
a. fo2
kOda.3.21 e-4 ni-4 u2
ji3
Nonce-word exist.
I see it(nonce).acc see
‘There is a nonceword. I saw it(the nonceword).’
b. gbele4.3
kOda.3.21 e-4 ni-4 e2
ji3
Nonce-word exist.
I see it(nonce).acc see
‘There is a nonceword. I saw it(the nonceword).’
• The antecedent does not have to be in the same utterance, nor nearby in the
discourse for this agreement to hold.
2
An interface model of phonologically determined agreement
Hannah Sande, UC Berkeley
2.2
WCCFL
March 2015
Phonological agreement between nouns and modifiers
• The same kind of agreement pattern holds between nouns and the final vowel of
adjectives that directly modify them (6).
(6) Noun-modifier phonological agreement1
a. [email protected] [email protected] éEla1.1 b. fu3 lelo1.2 éElO1.1
house new red
sponge new red
‘A new red house’
‘A new red sponge’
• I return to noun-modifier agreement in Section 5.2
3
Unsatisfying possible analyses
• A purely phonological analysis would involve stipulating that the pronoun and its
nominal antecedent are in correspondence, and ensure identity between the two
corresponding items (Sande, 2014).
– A phonological account alone won’t work because the nominal antecedent does
not have to appear within the same utterance, nor nearby in the discourse for
the phonological agreement to hold.
• A purely syntactic account could take one of two forms:
1. Phonological features are present in the syntax and available for copying during
morphosyntactic agreement processes.
2. Final vowels on nouns, and their agreeing pronoun vowels, are simply arbitrary
noun classes that coincidentally surface as entirely phonologically predictable.
• Option one defies accepted theoretical claims that syntax does not have access to
phonological information (Pullum and Zwicky, 1986, 1988).
• Option two predicts exceptions to the phonological predictability, and might predict
a default noun class for loan words or certain semantic categories.
– Since we find no exceptions and all pronoun agreement is predictable based
on the phonological features of the noun, this analysis fails to capture the
generalization that all noun-pronoun agreement is phonologically predictable.
4
An interface model of phonologically determined
agreement
• I propose a new approach to phonologically determined agreement that relies on
specific interactions between morphology and its interfaces.
• To account for the data, I propose a novel model of how ellipsis occurs at PF.
1
Word-internal ATR harmony influences the quality of the final vowel of the adjectives; however the
backness and rounding values of the final vowel are determined by the final vowel of the noun.
3
An interface model of phonologically determined agreement
Hannah Sande, UC Berkeley
WCCFL
March 2015
• The analysis in summary:
1. The agreement-controlling noun is present in the syntax, even when it does
not surface.
∗ Pronouns in Gu´ebie are in complementary distribution with the definite
marker, /-wa/, (7e).
∗ Like the definite marker, pronouns can occur with an overt noun, like the
‘we linguists’ construction in English, (7c).
∗ They can also license ellipsis of their nominal complement, (7d).
· The noun-pronoun construction differs from topic, focus, and definiteness in Gu´ebie.
(7) Distribution of definite markers and pronouns2
a.
b.
c.
sukulu1.1.3
sukulu-wa1.1.3.3
sukulu1.1.3 u3
‘school’
‘the school’
‘it school’
d.
e.
f.
u3
*sukulu-wa1.1.3.3 u3
*wa
‘it’ (the school)
‘it the school’
‘the (one/school)’
2. The morphology inserts an Agr(eement) node on the pronoun, and features of
the noun are copied to it.
3. The phonology, which applies at phase boundaries, can see the morphosyntactic features of the heads within that phase.
4. Phonological identity ensures that agreeing heads are phonologically similar.
5. Ellipsis of the noun occurs at PF and is licensed by overt phonological agreement between the noun and the pronoun.
4.1
The analysis: Syntactic structure to morphology
• The syntactic structure: The nominal antecedent of a pronoun is always present
in the syntax as complement to the pronoun D (cf. Elbourne 2001).
(8) Syntax of pronoun DPs in Gu´
ebie
DP
NP
D
{sukulu:N,E}
• The structre in (8) is identitcal to the sturucture of a definite DP. This follows
Postal (1966); Elbourne (2005); and Arkoh and Matthewson (2013) in unifying
determiners with pronouns.
• Morphology and phonology apply cyclically by phase (Marvin, 2002), and DP is a
phase (Svenonius, 2004).
• The morphological structure: An AGR-node is inserted on D, and the N feature
is copied to it from the noun (cf. Halle and Marantz 1994).
2
There are four contrastive tone heights in Gu´ebie, marked here with numbers 1-4 where 4 is high.
4
An interface model of phonologically determined agreement
Hannah Sande, UC Berkeley
WCCFL
March 2015
(9) Morphological agreement
DP
NP
{sukulu:N,E}
4.2
D
D {AGR:N}
The analysis: From morphology to phonology
• The phonological application: The morphological structure in (9) is linearized
via Distributed Morphology Linearization mechanisms (as laid out in Embick 2010).
• The morphological features associated with terminal nodes are preserved through
morphology and are available to phonology.
• I adapt a constraint-based approach, combining Agreement-by-Correspondence (Rose
and Walker 2004) with paradigm output-output faithfulness (Burzio 1994, Benua
1997, Kager 1999).
• The following widely used constraints accurately select the correct noun-pronoun
structures in Gu´ebie: [sukulu u], not *[sukulu e], *[sukule e].
– Ident-OO(N) (Benua, 1997): Heads that Agree in N must be phonologically
identical.
∗ This constraint is a stand-in for the combination of a Correspondence
constraint (Hansson, 2001; Rose and Walker, 2004) and an Ident constraint.
– Anchor-R (McCarthy and Prince, 1993): Segments at the right edge of agreeing heads correspond.
– RealizeMorph(eme) (Kurisu, 2001): Each morpheme has segmental content.
– *Struc(ture) (Prince and Smolensky, 1993): Assign one violation for each
output segment.
– Ident-IO (Prince and Smolensky, 1993): Assign one violation for each output
segment whose features differ from the corresponding input segment.
– Additional constraints (PeriphVowel, *i) account for the reduced number
of pronoun vowels compared to the full Gu´ebie vowel inventory.
(10) Ranking: Ident-OO(N), Anchor-R, Ident-IO, RealizeMorph *Struc
• For cases where a pronoun surface without a nominal complement, I posit that the
noun is present in the syntax but is elided at PF, [sukulu u] (cf. Merchant 2001;
Lasnik 2007).
• Constituents that can optionally be elided are marked with a feature E in the syntax
(Merchant, 2001).
• I propose a new model of ellipsis where the phonology has access to the E feature
of the noun, and the option of eliding the noun is determined via constraints.
5
An interface model of phonologically determined agreement
Hannah Sande, UC Berkeley
WCCFL
March 2015
– Faith-NoElide: For each form in an ellipsis paradigm, assign one violation for each output segment whose features differ from corresponding output
segments across the paradigm.
∗ This output-output paradigm correspondence constraint ensures that the
phrase containing the elided element be as similar to the non-elided output
as possible.
∗ For example, the elided form [sukulu u] must be faithful to the non-elided
[sukulu u].
(11) A constraint-based approach to ellipsis
{sukulu:N,E} {D:N}
a. sukulu u, u
Faith-NoE
b. sukulu u, Ø
c. sukulu s, s
d. sukulu e, e
e. sukule e, e
f. sukulu u, e
Id-IO
Id-OO(N)
Anchor
Realize
Max
*Struc
*
**!
*
*
*
*
6
6
6
6
6
6
8
7
8
8
8
8
*!
*!
*!
*!
• This analysis forces phonological agreement and provides the option of ellipsis at
PF simultaneously via constraints.
(12) An interface model of Gu´
ebie pronoun DP agreement
Syntax
Morphology
Phonology
DP
{sukulu:N,E} {AGR:N}
DP
NP
{sukulu:N,E}
5
D
NP
D
{sukulu:N,E}
{sukulu/sukulu}
[u]
D {AGR:N}
Extending the model to account for the entire Gu´
ebie
agreement system
5.1
Extending the model to human pronouns in Gu´
ebie
• Human pronouns in Gu´ebie do not follow the phonological agreement pattern that
all other nouns follow.
• They predictably take the forms /O/, singular, and /wa/, plural.
• I repeat the pronoun chart for Gu´ebie from (2) in (13) below.
(13) Human and non-human subject pronouns
Human
Non-human
Singular Plural
Singular Plural
1st e4
a3
1st –
–
2
2
2nd e
a
2nd –
–
3 3 3
3
3
3rd O
wa
3rd e ,@ ,u
i3 ,wa3
6
An interface model of phonologically determined agreement
Hannah Sande, UC Berkeley
WCCFL
March 2015
• The model described in Section (4) need be modified only slightly to account for
human pronouns in Gu´ebie.
• We saw that nouns are present in the syntax as complement to the pronoun, and
their features are copied to the pronoun via a morphological Agr node.
• Human nouns not only have a Noun feature which is copied to the pronoun, they
also have a [Person] feature (Richards, 2008; Van der Wal, 2015), as in Nudi ‘man,’
(14).
(14) Syntactic representation of human pronouns
DP
NP
D
{Nudi:N;Person:3;Sg;E}
• When features are copied from a human noun to the Agr node on the pronoun D,
Person, Number features are copied along with the Noun feature.
(15) Morphological agreement between human nouns and pronouns
DP
NP
{Nudi:N;Person:3;Sg;E}
D
D {AGR:N;Person:3;Sg}
• During Vocabulary Insertion, this bundle of features is spelled out as [O], as in (13).
(16) Phonological representation of human pronouns
{Nudi:N;Hum:3,Sg;E}
{AGR:N;Hum:3,Sg}
{Nudi/Nudi}
[O]
• We see that if certain semantic features of the noun (human) are copied to the
pronoun D via morphological agreement mechanisms, the pronoun is not subject
to phonological identity.
• Phonological identity is a last resort agreement strategy.
• This follows from Corbett (1991)’s generalization that when semantic and phonological criteria for determining noun class are at odds, the semantics takes precedence.
5.2
Extending the model to Gu´
ebie adjectives
• We saw that adjectives agree in final vowel with the nouns that they modify.
(17) Noun-modifier phonological agreement
a. [email protected] [email protected] éEla1.1 b. fu3 lelo1.2 éElO1.1
house new red
sponge new red
‘A new red house’
‘A new red sponge’
7
An interface model of phonologically determined agreement
Hannah Sande, UC Berkeley
WCCFL
March 2015
• We can derive this agreement in the same way as noun-pronoun agreement.
• Syntactically, nouns and the adjectives that directly modify them are present in the
same syntactic phase (DP).
• An AGR node is inserted on the adjective by the morphology.
• Features of the noun are copied to the adjective so that the adjective and noun are
in morphosyntactic agreement.
• The phonology ensures that agreeing heads (the noun and its adjectival modifiers)
are phonologically similar via the constraints discussed in Section 4.
– In the same way that pronouns license ellipsis of their nominal complement
(7d), adjectives agree with the head noun and license ellipsis of that noun.
(18) Overt agreement on adjectives
b.
a. [email protected] éEla1.1
new red
‘A new red one’ (house)
licenses ellipsis of the noun
lelo1.2 éElO1.1
new red
‘A new red one’ (sponge)
• Just like optional nominal ellipsis in noun-pronoun constructions, candidates are
evaluated in paradigms, with two forms in each paradigm: one where the noun is
elided and one where it is overt.
• A Faith-NoElide constraint ensures output-output paradigm faithfulness so that
the adjective agrees phonologically with the noun even when the noun is elided.
6
Extending the model to other phonologically determined agreement systems
6.1
Typological predictions
• The constraints presented in Section 4 account for the Gu´ebie data.
• They also predict the existence of phonologically determined agreement systems
slightly distinct from the one found in Gu´ebie.
• Predictions:
A. Any two elements within the same syntactic phase could surface in phonological agreement.
B. Phonologically corresponding segments will be edge-based or surface within
some prominent position in a word.
C. Any edge-aligned or prominent segment or suprasegmental can control agreement.
• Very few languages outside of Kru have described phonologically determined agreement systems; however, we see the above predictions born out in phonological
agreement systems crosslinguistically.
8
An interface model of phonologically determined agreement
Hannah Sande, UC Berkeley
6.2
WCCFL
March 2015
Other Kru languages
• A similar phonologically determined agreement system is present in related Kru
languages:
– Krahn, a Western Kru language (Bing, 1987)
– Godi´e, an Eastern Kru language (Marchese, 1986, 1988)
– Vata, an Eastern Kru language (Kaye, 1981; Marchese, 1979; Corbett, 1991)
(19) Phonological agreement across Kru
Gu´
ebie
Number of Agreeing Vowels
3
(Non-human) Pronoun-Noun Yes
Adjective-Noun
Yes
Demonstrative-Noun
N/A
Definite-Noun
No
Relative Pronoun-Noun
No
Krahn
3
Yes
No
No
No
No
Godi´
e
3
Yes
Yes
Yes
Yes
No
Vata
5
Yes
No
No
No
Yes
• The phonologically determined agreement systems in Krahn, Godi´e, and Vata all
closely resemble the Gu´ebie system except that a different set of elements agrees
with the noun in each language.
• Because all of the agreeing elements occur within the DP phase, each system above
is predicted by the proposed analysis (cf. Prediction A)
6.3
Bainuk
• Bainuk is a Western Atlantic language spoken in Senegal and Guinea (Sauvageot,
1967).
• Most nouns in Bainuk take one of 18 fixed noun class prefixes; however, there is a
class of prefixless nouns that triggers phonologically determined agreement.
• The first syllable of a prefixless noun, no matter its shape, surfaces as the agreement
marker on demonstratives, numerals, Wh-words, adjectives, and pronouns.
(20) Bainuk prefixless noun agreement
a. kata:ma-˜a
ka-nak-˜a b. dapOn
river
two
grass
‘two rivers’
‘long grass’
da-wuri
long
• Left-anchored correspondence as in Bainuk is predicted by the proposed analysis
(cf. Prediction B).
6.4
Abuq
• Abuq is a dialect of Arapesh spoken in Papua New Guinea (Nekitel, 1986).
• The final consonant of a noun triggers phonological agreement on demonstratives,
adjectives, and verbs in Abuq (Dobrin, 1995).
9
An interface model of phonologically determined agreement
Hannah Sande, UC Berkeley
WCCFL
March 2015
(21) Abuq phonological agreement (Nekitel 1986 cited in Dobrin 1995)
a. aleman afu-neri n-ahe’
man
good
went
‘a good man went’
b. almil aful-li l-ahe’
bird good went
‘a good bird went.
c. ihiaburuh afu-hi h-ahe’
butterfly good went
‘a good butterfly went.
• Traditionally there are 13 possible final consonants. Since contact with Tok Pisin
and other languages, words have been borrowed with other final consonants.
• Even in borrowed words with non-native segments, the final consonant of the noun
triggers agreement. This is clearly a phonologically-determined system.
(22) Borrowed words undergo phonological agreement
a. pater ara
priest this
‘This priest’
b. paip apa
pipe this
‘This pipe’
• The analysis proposed for Gu´ebie predicts a system like Abuq where edge-aligned
consonants (as opposed to vowels) control agreement (cf. Prediction C).
7
Conclusion
• Here I provide an initial description of the phonologically determined agreement
system of Gu´ebie (Kru, Niger-Congo).
• I provide an interface analysis where agreement arises through phonological identity
to output forms via morphological agreement mechanisms.
• The analysis includes a formal account of ellipsis via constraints at PF.
• I have shown that the proposed analysis predicts the attested cross-linguistic phonologically determined agreement systems (Corbett, 1991; Dobrin, 1995).
– I leave as a question for further research whether the proposed analysis could
serve as a model of gender and noun class systems more generally.
• I demonstrate that phonologically determined agreement systems can be modeled
without requiring phonological features to be present in syntax. Thus, I conclude
we can maintain that syntax is not sensitive to phonological features.
10
An interface model of phonologically determined agreement
Hannah Sande, UC Berkeley
WCCFL
March 2015
References
Arkoh, Ruby, and Lisa Matthewson. 2013. A familiar definite article in akan. Lingua
123:1–30.
Benua, Laura. 1997. Transderivational identity. Doctoral Dissertation, University of
Massachusetts, Amherst.
Bing, Janet. 1987. Phonologically conditioned agreement: Evidence from krahn. Current
approaches to African linguistics 4.
Corbett, Greville. 1991. Gender. cambridge textbooks in linguistics. Cambridge: Cambridge .
Dobrin, Lise M. 1995. Theoretical consequences of literal alliterative concord. In The
Proceedings of the 31st Regional Meeting of the Chicago Linguistic Society, volume 1,
127–142.
Elbourne, Paul. 2001. E-type anaphora as np-deletion. Natural Language Semantics
9:241–288.
Elbourne, Paul D. 2005. Situations and individuals, volume 90. MIT Press Cambridge.
Embick, David. 2010. Localism versus globalism in morphology and phonology, volume 60.
MIT Press.
Halle, Morris, and Alec Marantz. 1994. Some key features of distributed morphology.
MIT working papers in linguistics 21:88.
´
Hansson, Gunnar Olafur.
2001. Theoretical and typological issues in consonant harmony.
Doctoral Dissertation, University of California, Berkeley.
Kaye, Jonathan D. 1981. La s´election des formes pronominales en vata. Revue qu´ebecoise
de linguistique 11:117–135.
Kurisu, Kazutaka. 2001. The phonology of morpheme realization. Doctoral Dissertation,
University of California Santa Cruz.
Lasnik, Howard. 2007. On ellipsis: The pf approach to missing constituents. University
of Maryland working papers in linguistics 15:143–153.
Marchese, Lynell. 1979. Atlas linguistique kru. Abidjan: ILA.
Marchese, Lynell. 1986. Tense/aspect and the development of auxiliaries in kru languages.
Summer Institute of Linguistics.
Marchese, Lynell. 1988. 16 noun classes and agreement systems in kru: a historical
approach. Agreement in natural language: Approaches, theories, descriptions 323.
Marvin, Tatjana. 2002. Topics in the stress and syntax of words. Doctoral Dissertation,
Massachusetts Institute of Technology.
McCarthy, John J, and Alan Prince. 1993. Generalized alignment. Springer.
11
An interface model of phonologically determined agreement
Hannah Sande, UC Berkeley
WCCFL
March 2015
Merchant, Jason. 2001. The syntax of silence: Sluicing. Islands, and the Theory of .
Nekitel, Otto. 1986. A sketch of nominal concord in abu’(an arapesh language). Papers
in New Guinea Linguistics No. 24, Pacific Linguistics A70 .
Postal, Paul. 1966. On so-called pronouns in english. Monograph series on language and
linguistics 19:177–206.
Prince, Alan, and Paul Smolensky. 1993. 2004. Optimality Theory: Constraint interaction
in generative grammar .
Pullum, Geoffrey, and Arnold Zwicky. 1988. The syntax-phonology interface. In Linguistics: The Cambridge survey, ed. Newmeyer, 255–280. New York: The Cambridge
University Press.
Pullum, Geoffrey K, and Arnold M Zwicky. 1986. Phonological resolution of syntactic
feature conflict. Language 751–773.
Richards, Marc. 2008. Defective agree, case alternations, and the prominence of person.
In Scales, 137–161. Linguistische Arbeitsberichte.
Rose, Sharon, and Rachel Walker. 2004. A typology of consonant agreement as correspondence. Language 80:475–31.
Sande, Hannah. 2014. Pronoun-antecedent phonological agreement in Gu´ebie. UC Berkeley Phonology Lab Annual Report .
Sauvageot, Serge. 1967. Note sur la classification nominale en bainouk . Ed. du Centre
National de la Recherche Scientifique.
Svenonius, Peter. 2004. On the edge. In Peripheries, 259–287. Springer.
Van der Wal, Jenneke. 2015. Object clitics in comparative Bantu syntax. BLS 41.
12